GET Documentation¶

GET library¶

The GET library is the collection of CellML models which form the library of reusable modules utilised by the GET tools to create modular epithelial cell model descriptions. SED-ML will be used to provide descriptions of the simulation experiments to accompany the mathematical models in the library, to demonstrate specific instantiations of the models into simulation experiments. The library is available via both a Physiome Repository workspace and BitBucket. The Physiome Repository workspace should be considered the primary source and we maintain the clone on BitBucket to take advantage of the range of Mercurial management tools available on that platform. The ease with which we can do this demonstrates one of the advantages of using a distributed version control system like Mercurial.

The GET library currently consists of the core modules used by GET creator. The intention is to extend this library with a collection of curated CellML encodings of a range of epithelial transporters and whole cell models. The current collection of uncurated models can be seen in the renal nephron workspace.

GET creator¶

GET creator is the tool within the GET framework which focuses on the automated creation (and potentially editing in the future) of epithelial cell models via the assembly of constituent components from the GET library. In this manner, users are able to rapidly create customised epithelial cell models which capture the level of detail they require for their specific investigation. Through the incorporation of knowledge about the modules available in the library, GET creator is able to automatically define the glue which binds the generic modules together (i.e., summing all fluxes of a particular molecule through a given membrane).

We aim to infer model encoding requirements from annotations of the models contained in the GET library, but currently GET creator has that knowledge embedded directly in the code. By shifting this knowledge to the model annotations, our approach will be extensible to any sufficiently annotated model that might be added to the library in the future. This will become important as the scope of the library extends beyond the current core models.

GET creator is designed as a library with a clearly defined programming interface. The code repository contains a demonstration application which makes use of this library to start the process of creating the model described by Latta et al (1984). While this test application is complete, the generated CellML model is not yet complete.

In addition to creating the CellML model configured for the users specific requirements, GET creator will also be able to customise generic simulation experiment descriptions from the GET library that are deemed suitable for the model being constructed. While certain basic simulation experiments that might be suitable can be inferred from the library models being integrated into the new cell model, further guidance from the user will result in a more comprehensive suite of simulation experiments being generated. In this manner, we envision the incorporation of some of the functional curation concepts developed for cardiac electrophysiology models could be transfered to epithelial cell transport models.

GET simulator¶

GET simulator is the simulation engine for the GET framework. Underneath, GET simulator makes use of CSim, a generic CellML simulation tool with both a command line client and a reusable library.

While whole epithelial cell models can be encoded in CellML and those models can be simulated by standard CellML-capable simulation tools, in most cases that will not be the full experiment the user desires to run. Instead, the method described by Latta et al (1984) has been implemented in GET simulator in order to solve for mass and charge conservation. The idea is that GET simulator makes use of CSim to evaluate parts of the model while following the Latta et al (1984) algorithm for solving the current state of the whole cell.

As mentioned above, SED-ML is used to encode the descriptions of the simulation experiments to be executed. SED-ML makes use of the Kinetic Simulation Algorithm Ontology (KiSAO) to describe the numerical method to be used when executing a particular numerical simulation task. KiSAO does not currently contain suitable terms to describe the simulation algortihm implemented by GET simulator. So we are making use of a proposed extension for SED-ML for the inclusion of custom simulation algorithms. In this manner, we are able to encode our epithelial cell simulation experiments in generic SED-ML with all the advantages that that offers, but still have a method by which to inform other simulation engines that a specific algorithm is required. A suitable fall-back algortihm can also be defined such that simulation engines are able to perform the default integration of the cell model as mentioned above.

GET web application¶

The GET web application is an attempt to provide a web-based drag-and-drop interface for assembling epithelial cell models. This interface makes use of the GET library and GET creator tools above via web services provided by the GET model server. You can see a screenshot of the current state of the interface embedded in the Physiome Modeller interface here.

GET model server¶

The GET model server is a prototype web server providing access to the GET framework via standard web services. These services are specific to the GET framework, but as the project develops common tools will be extracted out as proposed features to be implemented as part of the software which runs the Physiome Repository.

Overview¶

The three aspects of the Physiome Modeller user interface are:

• the tree view on the left, allowing the user to navigate through the spatial scales;
• the centre panel, which displays the actual content; and
• the right panel shows the interactive graphical rendering of the model – i.e., the same as the nephron graphic in the original nephron browser.

In the example shown below, you see the “human” spatial scale where you get the search results for all data in the Physiome Repository matching “circulatory system” in the content pane and a whole body to play with in the graphical view.

Since we are using our fully fledged visualisation toolkit, we have access to all sorts of useful functionality. In this screen shot we have displayed some of the original images used to create this computational model (in this case from the visible human data set).

and here showing that as the user “zooms” in to the model, the skin has disappeared to allow uninterrupted views of the components of the human circulatory system.

In the Physiome Modeller, the user is able to navigate through the spatial scales, for the renal system only, either “automatically” by zooming in on the appropriate region of the graphical display; or by selecting the desired scale in the tree view on the left panel. Here we have shifted to the kidney (organ spatial scale) and can now see all search results from the Physiome Repository related to the kidney (in the centre panel) and the graphical rendering shows the larger vessels of the renal vasculature.

Zooming in further, the user drops down to the sub-organ spatial scale and is presented with the nephron model (from the previous nephron browser) and accompanying nephron search results from the repository.

Further zooming now drops the user down to the cell spatial scale. In the demonstration shown below, the user is presented with a certain epithelial cell model in the graphical rendering pane, but that has been hidden in this screen shot to allow a better view of the GET web application now being embedded inside the Physiome Modeller user interface. While not actually integrated into one web application, it does at least present a common look and feel across the tools and show where things are heading. The user is also able to easily navigate back up to the higher spatial scales, so that once we get the simulation capabilities all hooked up you could imagine how a user could “edit” their cell model in this view and then run whole nephron or kidney simulations and quick access the simulation results all in one tool.

When available for the current model being displayed, it is also possible for the user to browse an embedded version of the CellML reference description (described above), as shown below. Again, the model visualisation panel has been hidden.

General¶

Within sections¶